Customize with AI

Volt is unusually friendly to AI coding tools (Claude Code, Cursor, Copilot, …) — and that's not an accident. The qualities that make it small make it easy for an AI to understand and change correctly.

Why it works so well

• One readable file. The entire UI library is ~260 lines of plain JS — an AI can hold all of it in context and reason about the whole framework, not a slice of a giant ecosystem.
• No build step. The AI edits a file, you save, it hot-reloads. There's no bundler/transpiler config to get wrong, and no opaque error surface between the code and the result.
• Plain files, not a database. Your app is server.js, public/app.js, views/, and .env — readable, diffable, version-controlled. There's no hidden config in a DB (the WordPress problem) for an AI to be blind to.
• Safe by construction. Volt interpolations render as escaped text nodes, so an AI can't accidentally introduce an XSS hole by templating user data.

How to do it

Open the app in your AI editor, point it at the relevant files, and ask. Give it public/volt.js plus the file you're changing, and mention the API (signal / computed / el / html / mount). Then run npm run dev and watch it hot-reload.

npm run dev      # keep it running; AI edits hot-reload live

Prompts that just work

• "Add a priority field to tasks — a low/med/high dropdown — and sort the list by it."
• "Add a dark/light theme toggle stored in localStorage."
• "Add pagination to the notes list, 20 per page."
• "When a new item is added, email me a summary using the mailer add-on."
• "Turn on realtime so the list updates live across tabs." (the AI enables it via --edit + the realtime add-on)

Tip: have the AI run the app and verify in a browser, not just write code. No build step means the feedback loop is seconds.

Volt itself — and this docs content — are authored this way.